We present 16 GHz (1.9 cm) deep radio continuum observations made with the Arcminute Microkelvin Imager (AMI) of a sample of low-mass young stars driving jets. We combine these new data with archival information from an extensive literature search to examine spectral energy distributions (SEDs) for each source and calculate both the radio and sub-mm spectral indices in two different scenarios: (1) fixing the dust temperature (T_d) according to evolutionary class; and (2) allowing T_d to vary. We use the results of this analysis to place constraints on the physical mechanisms responsible for the radio emission. From AMI data alone, as well as from model fitting to the full SED in both scenarios, we find that 80 per cent of the objects in this sample have spectral indices consistent with free–free emission. We find an average spectral index in both T_d scenarios, consistent with free–free emission. We examine correlations of the radio luminosity with bolometric luminosity, envelope mass and outflow force, and find that these data are consistent with the strong correlation with envelope mass seen in lower luminosity samples. We examine the errors associated with determining the radio luminosity and find that the dominant source of error is the uncertainty on the opacity index, β. We examine the SEDs for variability in these young objects, and find evidence for possible radio flare events in the histories of L1551 IRS 5 and Serpens SMM 1.

AMI radio continuum observations of young stellar objects with known outflows

In this paper, we propose a clustering approach for solving the problem of reconstructing cross-cut shredded documents. This problem is important in the field of forensic science. Unlike other clustering approaches which are applied as a preprocessing step before the actual reconstruction algorithms, our clustering approach is part of the reconstruction process itself. We define a new cost function which mainly relies on black pixels to measure the cost of pairing two shreds together. The reconstruction algorithm creates multiple clusters which grow by adding additional shreds based on the cost function. Adding a shred may result in merging two or more clusters to produce a larger cluster. We, also, propose a way to involve the user in the reconstruction process. We compare our approach with a recent proposal and conclude that our approach gives better solutions in less time.

An alternative clustering approach for reconstructing cross cut shredded text documents

The ubiquitous availability of big data and computational resources has accelerated the development and adoption of artificial intelligence methods in many application domains albeit with many challenges. This paper proposes a novel safety assurance model for aviation systems. The model exploits the ensemble of machine learning and deep learning algorithms to improve the prediction of risks and risks’ severity in aviation systems. The diversity score of base classifiers is used to fine-tune the ensemble performance. The model is trained and evaluated using twelve years of data from the Aviation Safety Reporting System (ASRS) database. We demonstrate the superiority of our model using numerical evaluation criteria, Precision, Recall, and F-Score. Finally, we conclude and give research directions for the future.

An Ensemble Machine and Deep Learning Model for Risk Prediction in Aviation Systems

Web service is accepted by all communities of services due to its interoperability, flexibility of design and architecture. Improving performance is the key research point for developers and researchers. We propose to design, develop and implement a prototype research electronic automated teller machine web service to study the loads, performance and scalability of the service. The service is based on Visual Studio .NET framework. It is developed by using C# as programming language, Internet Information Service (Version: 2005) as web server and, Microsoft Structured Query Language as database server. To study its different aspects the service has been tested by deploying it on Mercury LoadRunner (Version 8.1). In this paper we will present the architecture, testing procedure, results of performance testing as well as results of statistical analysis on recorded data of the service.

Investigations on implementation of e-ATM Web Services based on .NET technique

The World Wide Web is a large distributed digital information space. The ability to search and retrieve information from the Web efficiently and effectively is an enabling technology for realizing its full potential. Information Retrieval IR plays an important role in search engines. Today's most advanced engines use the keyword-based "bag of words" paradigm, which has inherent disadvantages. Organizing web search results into clusters facilitates the user's quick browsing of search results. Traditional clustering techniques are inadequate because they do not generate clusters with highly readable names. This paper proposes an approach for web search results in clustering based on a phrase based clustering algorithm. It is an alternative to a single ordered result of search engines. This approach presents a list of clusters to the user. Experimental results verify the method s feasibility and effectiveness.

A Roadmap to Integrate Document Clustering in Information Retrieval

Signal Processing and Information Technology

Semantic annotation represents a metadata of the document based on domain ontology. The purpose of this paper is to develop a ranking algorithm for semantic document annotation and to evaluate its performance in the Semantic Web SW application. The evaluation is to compare the ranking algorithm against other algorithms. For the evaluation purpose, all the algorithms are applied into the SW application. The SW application is a research prototype retrieval engine, PicoDoc. The system framework of PicoDoc is based on OCAS2008 ontology. During the experimentation stage, a real-life dataset from news article corpus of ABC and BBC websites are selected. The experiment shows promising results in retrieving related information using the ranking algorithm.

Ranking Algorithm for Semantic Document Annotations

The aim of this research is to analyse the performance of six different classifiers, which are κ-Nearest Neighbours (kNN), Naive Bayes, Random Tree, J48 Decision Tree, Random Forest Tree and Sequential Minimal Optimisation (SMO), using aircraft databases and optimize their cost parameter for better accuracy. The six algorithms are implemented to classify aircraft type and its country of origin using a Waikato Environment for Knowledge Analysis (WEKA) workbench. Additionally, we report our parameter optimisation results for SMO by varying the cost parameters to obtain the optimum result. It is observed that in both classifications, SMO with linear kernel obtained the best performance as compared to the other classifiers in terms of classification accuracy, which is 100%.

Performance comparison of machine learning classifiers on aircraft databases

Learning programming language of Java is a basic part of the Computer Science and Engineering curriculum. Specific Java compiler is a requirement for writing and convert the writing code to executable format. However, some local installed Java compiler is suffering from compatibility, portability and storage space issues. These issues sometimes affect student-learning interest and slow down the learning process. This paper is directed toward the solution for such problems, which offers a new programming assignment submission system with cloud-based Java compiler and is known as CheckMyCode. Leveraging cloud-computing technology in terms of its availability, prevalence and affordability, CheckMyCode implements Java cloud-based programming compiler as a part of the assignment management system. CheckMyCode system is a cloud-based system that allows both main users, which are a lecturer and student to access the system via a browser on PC or smart devices. Modules of submission assignment system with cloud compiler allow lecturer and student to manage Java programming task in one platform. A framework, system module, main user and feature of CheckMyCode are presented. Also, taking into account are the future study/direction and new enhancement of CheckMyCode.

CheckMyCode: Assignment Submission System with Cloud-Based Java Compiler

With the support of semantic annotation and domain ontology, semantic web is able to assist people in querying rich documents. However, generating queried semantic documents without ranking them in a right order is ineffective. In this paper, we are extending FF-ICF algorithm with the concept spreading. For experimentation, this algorithm is applied into a research prototype retrieval engine, PicoDoc. The PicoDoc system uses corpus that has pre-annotated documents as its data reference to run query against, based on real-life dataset from ABC and BBC news article corpus. The corpus is based on OCAS2008 ontology. The experiment shows a modified FFICF-related spread concept yields promising results in retrieving related information.

Syarifah Bahiyah Rahayu

Papers

Signal Processing and Information Technology

Ranking Algorithm for Semantic Document Annotations

Performance comparison of machine learning classifiers on aircraft databases

CheckMyCode: Assignment Submission System with Cloud-Based Java Compiler

Semantic Document Annotation Ranking Model